Andrés Rodríguez-Pose
, Yannis Psycharis and
Vassilis Tselios
Liberals, socialists, and pork-barrel politics
in Greece
Article (Accepted version)
(Refereed)
Original citation:
Rodríguez-Pose, Andrés, Psycharis, Yiannis and Tselios, Vassilis (2016) Liberals, socialists, and
pork-barrel politics in Greece. Environment and Planning A, 48 (8). pp. 1473-1492. ISSN 0308-518X
DOI: 10.1177/0308518X16646372
© 2016 The Authors
This version available at: http://eprints.lse.ac.uk/67541/
Available in LSE Research Online: August 2016
LSE has developed LSE Research Online so that users may access research output of the School. Copyright © and Moral Rights for the papers on this site are retained by the individual authors and/or other copyright owners. Users may download and/or print one copy of any article(s) in LSE Research Online to facilitate their private study or for non-commercial research. You may not engage in further distribution of the material or use it for any profit-making activities or any commercial gain. You may freely distribute the URL (http://eprints.lse.ac.uk) of the LSE Research Online website.
This document is the author’s final accepted version of the journal article. There may be differences between this version and the published version. You are advised to consult the publisher’s version if you wish to cite from it.
1
Liberals, Socialists, and pork-barrel politics in Greece
byAndrés Rodríguez-Pose*, Yannis Psycharis** and
Vassilis Tselios***
Abstract
This paper analyses the role of pork-barrel politics in the allocation of public investment expenditures in Greece. It proposes a model which explicitly relates the allocation of public investment to electoral results using a unique dataset covering the period from the restoration of democracy in 1974 until 2009, just before the Great Recession that radically transformed the political panorama of the country. The analysis includes ten legislative periods marked by governments of the two parties that dominated the political arena in Greece: the Liberal and the Socialist Party. The results show that Socialist and re-elected governments applied more expansionary fiscal policies relative to Liberals. The two main parties also used different tactics when it came to pork-barrelling: while the Socialists when in government rewarded/groomed their electoral fiefs, the Liberals invested in areas controlled by the opposition to win over new votes or seats.
Keywords: Public investment, pork-barrel politics, elections, regional policy, Greece JEL classification: P16, R1, R12, R42, R58, H54
Authors’ addresses:
*Department of Geography and Environment London School of Economics
Houghton St
London WC2A 2AE, UK Tel: +44 (0)207 955 7971 Fax: +44 (0)207 955 7412
E-mail: a.rodriguez-pose@lse.ac.uk
**Department of Economic and Regional Development
Panteion University of Social and Political Sciences 136 Syngrou Avenue 176 71 Athens, Greece Tel: +30 210 9201722 Fax: +30 210 9232979 E-mail: psycharis@panteion.gr
*** Department of Planning and Regional Development University of Thessaly Pedion Areos Volos, 383 34, Greece Tel: +30 24210 74451 Fax: +30 24210 74380 E-mail: vtselios@prd.uth.gr
2
1. Introduction
The analysis of the territorial allocation of public expenditure is an area of research with a long tradition. At the crossroads between public economics, political science, and political geography, the literature on the implications of pork-barrel politics has increasingly become more theoretically and empirically sophisticated (Golden and Min 2013). Different empirical studies offer a variety of outcomes and interpretations, depending on the discipline of origin, the approach adopted, or the individual countries or groups of countries covered.
One of the drivers of this type of research is the realisation that government decisions about the geographical allocation of public spending are not always driven only by ‘objective’ socio-economic criteria, such as efficiency and/or equity, as the normative theory of public finance would claim (Buchanan 1950), but also by political considerations (Oates 1972; Bennett 1980). And in few countries has the link between pork-barrel politics and economic outcomes been more under the spotlight than in Greece, making it a particularly interesting – and, until now, neglected – case study to analyse for a number of reasons.
First, pork-barrelling in public investment is always prominent in countries with strong clientelistic and/or nepotistic networks. In Greece the provision of collective goods, such as bridges and highways or schools and hospitals, has often been regarded as a perk to pay back voters for their electoral support. Second, Greece being a highly centralised country, the decisions about how to allocate public projects – regardless of whether the actual expenditure takes place at the national or at the local level – can often be traced back to national elections. Third, the high level of Greek fiscal centralisation and the lack of a specific formula for the regional allocation of public investment leaves plenty of room for political bargaining – and, hence, for pork-barrelling – about the territorial allocation of public expenditure. Fourth, regional development was one of the avowed policy priorities of successive Greek national governments throughout the period of analysis. Infrastructure investment was used as the main mechanism for upgrading the competitiveness and cohesion of regional economies and was the target of a high proportion of the projects co-funded through EU structural assistance. Fifth, Greece organized the 2004 Olympic Games. The preparation of the Games meant that considerable financial resources were channelled towards infrastructure investment of dubious sustainability. In addition, the need to meet strict deadlines in the preparation for the Games and the implementation of parallel programmes to the Olympic Games, such as the so-called Greece 2004 – financing athletic and other types of infrastructures in every single municipality across the country –is considered to have allowed plenty of opportunities for pork-barrel politics.
The specific institutional conditions of Greece and the combination of factors outlined above hint that pork-barrel politics may have contributed to distorting regional development policy priorities and undermining the programmatic targeting towards the less well-off regions of the country. In addition, it might have contributed to a macroeconomic situation characterised by high public spending, growing deficits, and galloping debt.
This paper examines the extent to which political factors (votes, seats in parliament, and the party in office) have shaped expenditure on local public goods across the 51 Greek NUTS III regions (nomoi) between the return to democracy in 1974 and the outbreak of the crisis in 2009. The analysis considers the total volume of public investment expenditures by political period. Public investment expenditures constitute part of national budget and include national funds along with the structural assistance from the EU to finance public infrastructures in the country. Data have been disaggregated at NUTS III regional level, where administrative
3
divisions coincide with – with the exceptions of Athens and Thessaloniki – political constituencies.
The paper is structured as follows. Following this introduction, section 2 discusses scholarly approaches to pork-barrelling and highlights the key features of the Greek political setting. Section 3 presents the data, the variables, and the econometric model. Section 4 discusses the regression results, while section 5 concludes.
2. Pork-barrel politics and public investment
2.1 Politics and public investment
Theories of distributive politics frequently pay particular attention to the processes behind pork-barrel politics. Electoral considerations often play a key role in the allocation of public expenditure by incumbent governments. Rather than expenditure following ‘objective’ socio-economic criteria, in contexts where pork-barrelling is pervasive, electoral considerations and the expectations for and results of electoral contests may become fundamental drivers behind the territorial allocation of public expenditure. Incumbent governments, considered as profit maximisers, may allocate public resources with the aim of extracting the highest electoral benefit, rather than pursuing equity and/or efficiency.
Scholarly research provides ample evidence of the presence of pork-barrel politics. There are numerous examples in Europe. Limosani and Navarra (2001) find that in pre-election yeas local policy-makers in Italy have had a tendency to increase investment outlays beyond their standard growth rate. Dahlberg and Johansson (2002) demonstrate that in Sweden incumbent governments have used grant programmes to win votes. Celbis, de Crombrugghe and Muysken (2014) suggest that political bias has been present in the allocation decisions of regional transportation and communication public investments in Turkey.
Pork-barrelling is also widespread as a political practice in the Americas. In the United States Primo and Snyder (2010), for example, show that those areas of the US where a party clearly dominates, attract less attention in terms of distributive spending, because of a decreased incentive for individual legislators to secure a ‘personal vote’ via local projects. More examples of pork-barrelling have been described by Chen (2010) in the case of New York, who proves that the electoral geography of legislative districts affects pork-barrelling under bicameralism. He indicates that greater electoral fragmentation has a negative effect on pork-barrelling. Stokes (2005) depicts how in Argentina political machines (or clientelistic parties) mobilise electoral support by trading particularistic benefits to voters in exchange for their votes. Similarly, Costa-i-Font, Rodriguez-Oreggia, and Lunapla (2003) find a positive relationship between the regional allocation of public investment and past support for the central ruling party in Mexico.
Not all research, however, indicates that pork-barrelling is pervasive. De La Calle and Orriols (2010) for example, using as a case study the expansion of the underground network in Madrid, show that, although governments are often tempted to follow vote-seeking strategies, they cannot deviate too much from an efficiency-based allocation of public resources. Luca and Rodríguez-Pose (2015) establish that, although politics in Turkey plays a non-negligible role in influencing public investment allocations, the magnitude of pork-barrelling is relatively low in comparison to the role played by socioeconomic factors when determining how to allocate public investment across Turkish regions.
4
In pork-barrel environments, the specific allocation of public funding frequently depends on the electoral prospects of the ruling party in specific constituencies. Political constituencies can be divided accordingly into three types (Johnston 1977): a) ‘hopeless’ areas, where the ruling party has a very low level of electoral support and little hope of ever winning more support; b) ‘safe’ areas, where the ruling party has a high level of popular support and its victory in various electoral contests is virtually guaranteed; and c) ‘marginal’ areas, where the ruling party either has a slight majority or its opponents have a slight lead over it. In such a categorization, governments will prioritise, depending on their main objectives, either rewarding existing electoral support or gaining new support.
Two contesting models have been put forward in the scientific literature: the ‘loyal voter’ and the ‘swing voter’ model. The ‘loyal voter’ model assumes that governments allocate public funds to reward constituencies where their core supporters live (Cox and McCubbins 1986; Golden and Picci 2008). The ‘swing voter’ model assumes, by contrast, that governments prefer to splash public investment in those regions where the biggest electoral gains can be achieved (Lindbeck and Weibull 1987; Dixit and Londregan 1996; Weghorst and Lindberg 2013; Golden and Picci 2008). Whereas in the former model incumbent governments reward the past electoral support of their voters, in the latter they aim to buy-off votes by prioritizing expenditures to swing regions. Ansolabehere and Snyder (2006: 549), in an attempt to bridge this dichotomy, argue that it may also be possible that parties chose to target both loyal and swing areas simultaneously.
2.2 Greece, public investment, and pork-barrel politics
There is a widespread popular perception that politics plays an important role in the regional distribution of public spending, in general, and public investment, in particular, in Greece. It is not infrequent to read in the local press news about politicians boasting about their capacity to influence and ‘deliver’ investments in, say, roads or schools for their constituencies. Political arguments make the public investment budget – an important section of the national budget – one of the most politicised instruments of government policy in the country (Psycharis 1990). While more ‘objective’ socio-economic criteria determine the bulk of public expenditure, this part of the national budget – which was introduced in 1952, has been the mechanism channelling EU structural funds to Greece since 1982, and is today under the authority of the Ministry of Development – remains however extremely vulnerable to political wrangling and pork-barrelling. Ministers, mayors, members of Parliament, and other politicians vie to carve a share of the public investment budget for their constituencies (Psycharis 2008).
However, the popular view of a polity riddled by pork-barrelling has not really been confirmed by the scholarly literature. The number of studies on distributive politics in Greece remains limited. The few that exist generally validate the presence of pork-barrel politics in the regional allocation of public funds, but the results are extremely sensitive to the different approaches adopted. Lambrinidis, Psycharis and Rovolis (2005), for example, show that the percentage of votes in a prefecture in favour of the governing party has not been a fundamental driving force in the regional allocation of public investment in infrastructure across prefectures. Rodríguez-Pose, Psycharis, and Tselios (2012), by contrast, demonstrate that differences in votes between the governing and the main opposition party in each prefecture determine the territorial allocation of public investment, which, in turn, influences growth rates across Greek prefectures. Psycharis (2008) finds evidence that politicians in key positions in government have funnelled public spending to their constituencies and/or regions
5
of origin. As a result, the ways through which distributive politics influence the allocation of public investment are still open to investigation.
This paper aims to supersede previous studies by proposing a model using the total volume of public investment – rather than public investment per capita (as in Rodríguez-Pose, Psycharis, and Tselios 2015) – by political period as dependent variable, in order to examine how politics affects the regional allocation of public investments. The analysis controls for a series of variables that, under normal circumstances, would have affected the territorial distribution of public investment. We specifically focus on how public investment is allocated in order to provide different types of public goods and collective services to people in different areas of the country.
3. Data, Variables and Empirical specification: the Greek context
The main aim of the paper is to assess the impact of political factors on the distribution of total public investment expenditures across 51 Greek regions (NUTS III level) over 10 political periods, paying particular attention to the behaviour of different political parties with regards to pork-barrelling when in office. The choice of spatial level and political periods of analysis is not casual.
The spatial level of analysis is NUTS III regions. NUTS III regions largely coincide with the Greek political constituencies. In a Greek Parliament of 300 seats, 288 members of Parliament (MPs) are elected directly in open lists on the basis of constituency votes. The remaining 12 MPs (i.e. the so called national deputies or ‘Epikrateias’) are elected from closed lists proposed by political parties. The allocation of these MPs to different political parties is proportional to the political power of each party after the elections (Rodríguez-Pose, Psycharis, and Tselios 2015). The 288 directly-elected MPs were elected in 56 constituencies: 49 of these constituencies coincide with NUTS III regions. The only exceptions concern the regions of Attiki and Thessaloniki. Attiki is divided into five electoral constituencies (Athens A, Athens B, Piraeus A, Piraeus B, and Rest of Attiki). Thessaloniki, in turn, is divided into two electoral regions (Thessaloniki A and Thessaloniki B). The coincidence between NUTS III regions and electoral constituencies makes these regions the best unit of analysis for the link between politics and the territorial distribution of public investment.
The political periods are determined by the party in office following successive national elections between 17 November 1974 and 4 October 2009. During the period considered, two parties dominated the political arena in Greece: New Democracy (ND), also known as the Liberal Party, and the Panhellenic Socialist Movement Party (PASOK), also known as the Socialist Party. These parties alternated in government during the 10 political periods considered in the analysis.1 Table 1 displays a) the dates of the national Greek elections, b) which party was in office, and c) the changes in governing party. ND was the governing party in the periods 1975-1977, 1978-1981, 1990-1993, 2004-2007, and 2008-2009, while PASOK was in office during the periods 1982-1985, 1986-1989, 1994-1996, 1997-1999, and 2000-2003. During the period of analysis each party was in office for a total of 5 political periods. Changes in power took place in 1975 (from a junta to ND), in 1982 (from ND to PASOK), in 1990 (PASOK to ND), in 1994 (ND to PASOK), and in 2004 (PASOK to ND). Incumbent
1 We exclude the three short-lived governments during the period 1989-1990. Between 1989 and 1990 Greece
lived a rare – at least until the outbreak of the Great Recession – period of political instability. Three different governments replaced each other in quick succession: a coalition between the conservatives of New Democracy and the Left (03/07/89 - 12/10/1989); a caretaker government (12/10/89 - 23/11/89); and a ‘national unity’ government in which New Democracy shared power with both PASOK and the Left (23/11/1989 – 11/04/1990).
6
governments were re-elected five times, 1978 (ND), 1986 (PASOK), 1997 (PASOK), 2000 (PASOK), and 2008 (ND).
Insert Table 1 around here
Data on public investment have been provided by the Ministry of Development which is responsible for the Public Investment Programme of the country. Public investment expenditures include investment expenditures in agriculture, forestry, and fishing; manufacturing; energy, mining, and handicraft; transport infrastructures; tourism and modern culture; education and research; housing and environment; health and public welfare; water supply and sewage facilities; devolved public expenditures to regions and prefectures; special infrastructure works; miscellaneous and administrative expenditures; and Olympic Games infrastructure projects. EU structural assistance to Greece, the national contribution to EU co-financed projects, and projects of purely national interest and funding are all included in these data (Rodríguez-Pose, Psycharis, and Tselios 2015). For each political period, we calculated the total volume of public investment expenditures. Although the avowed aim of Greek public expenditure has always been – regardless of the political orientation of the party in office – that of achieving greater equity and/or efficiency, our hypothesis is that the territorial distribution of funds may have been affected by political factors, as public investment in Greece has always been open to political discretion. These expenditures may therefore capture pork-barrelling. Figure 1 shows the average regional distribution of the total volume of public investment expenditures between 1975 and 2009.
Insert Figure 1 around here
In particular, we formulate and empirically test three hypotheses regarding potential political influences in the allocation of public investment across Greek regions. First, we analyse the extent to which differences between parties in political power in a given region determine public investment expenditures in the region. Second, we assess whether the re-election of a party leads to greater pork-barrelling than a first-time election. Third, we examine whether public investment expenditures differ depending on the orientation of the party in office (i.e. between the Liberal and the Socialist Party). We use the following simple model:
𝐼𝑛𝑣𝑖𝑡 = 𝑓(𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡, 𝐶ℎ𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡, 𝑃𝑜𝑙𝑖𝑡𝑃𝑜𝑤𝑒𝑟𝑖𝑡, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖𝑡) (1) where Invit is the natural logarithm of the total volume of public investment expenditures in
region i (i = 1, 2, …, 51) at political period t (t = 1, 2, …, 10)2, GovPartyt is a dummy
variable coding whether the governing party at period t was ND (t = 1, 2, 5, 9, and 10) or PASOK (t = 3, 4, 6, 7, and 8) (base category), ChGovPartyt is a dummy variable indicating
whether the governing party at period t had been re-elected (t = 2, 4, 7, 8, and 10) or changed (t = 1, 3, 5, 6, and 9) (base category), PolitPowerit is a vector of regional political power
variables (i.e. pork-barrel politics variables) in region i and at period t, and Controlit is a
vector of control variables in region i and at period t.
2
The dependent variable adopts a logarithmic form, because the distribution of the total volume of public investment expenditures is skewed.
7
We expect differences in regional political power – following the ‘loyal’ vs. ‘swing’ voter model – to affect the allocation of public investment (Hypothesis 1). We do not anticipate differences in pork-barrelling behaviour between the Liberals and the Socialists (Hypothesis
2), as corruption is widespread in Greek society (European Commission 2014). Finally, we
consider that re-elected governments, that are already familiar with the levers of power, will resort to pork-barrelling in greater measure than first-time elected governments (Hypothesis
3). { 𝐼𝑛𝑣𝑖𝑡 = 𝑓(𝐶ℎ𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡, 𝑃𝑜𝑙𝑖𝑡𝑃𝑜𝑤𝑒𝑟𝑖𝑡, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖𝑡) 𝑖𝑓 𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡 = ′𝑁𝐷′ 𝐼𝑛𝑣𝑖𝑡 = 𝑓(𝐶ℎ𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡, 𝑃𝑜𝑙𝑖𝑡𝑃𝑜𝑤𝑒𝑟𝑖𝑡, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖𝑡) 𝑖𝑓 𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡= ′𝑃𝐴𝑆𝑂𝐾′ (2) and { 𝐼𝑛𝑣𝑖𝑡 = 𝑓(𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡, 𝑃𝑜𝑙𝑖𝑡𝑃𝑜𝑤𝑒𝑟𝑖𝑡, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖𝑡) 𝑖𝑓 𝐶ℎ𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡 = ′𝐶ℎ𝑎𝑛𝑔𝑒′ 𝐼𝑛𝑣𝑖𝑡 = 𝑓(𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡, 𝑃𝑜𝑙𝑖𝑡𝑃𝑜𝑤𝑒𝑟𝑖𝑡, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖𝑡) 𝑖𝑓 𝐶ℎ𝐺𝑜𝑣𝑃𝑎𝑟𝑡𝑦𝑡= ′𝑁𝑜 𝐶ℎ𝑎𝑛𝑔𝑒′ (3) Equation (1) tests Hypothesis 1, Equation (2) tests Hypothesis 2, and Equation (3) Hypothesis
3.
The regional political power variables are depicted by the percentage votes and seats (MPs) for each party in each region. More specifically, the regional political power variables include the regional political power of the Liberal Party, the Socialist Party, and of minority parties, measured as:
the percentage votes (seats) in the region in favour of ND (%ND – Votes (Seats));
the percentage votes (seats) in the region in favour of PASOK (%PASOK – Votes
(Seats)), and;
the percentage votes (seats) in the region in favour of minority parties (e.g. the Greek Communist(s) Parties) (%Rest – Votes (Seats)).
The regional political power variables also include the (‘absolute’) regional political power of the governing party and the regional political power of the governing party relative to the main opposition party. This latter variable is known as the ‘relative’ political power of the governing party (Rodríguez-Pose, Psycharis, and Tselios 2012). The variables, respectively, are measured as:
the percentage votes (seats) in the region in favour of the governing party (%GovParty – Votes (Seats));3 and
the difference in the percentage votes (seats) in the region in favour of the governing party (Dif%GovParty – Votes (Seats)).4
The value of the relative political power variable for a particular region can be negative in those cases where the percentage share of votes or seats in favour of the governing party (ND or PASOK) is lower than the percentage share of votes or seats in favour of the main opposition party (PASOK or ND, respectively) (Rodríguez-Pose, Psycharis, and Tselios 2012).
Finally, the regional political power variables include the political power of single-seat constituencies. This variable is measured by a dummy variable (Single-seat constituency) coding whether a region returns only one MP to Athens (i.e. single-member) or more (i.e.
3 Calculated as ‘%GovParty – Votes (Seats)’ = ‘%ND – Votes (Seats)’ x ‘GovParty’ + ‘%PASOK – Votes
(Seats)’ x (1 – ‘GovParty’).
4 Calculated as ‘Dif%GovParty – Votes (Seats)’ = (‘%ND – Votes (Seats)’ – ‘%PASOK – Votes (Seats)’) x
8
multi-member) (base category). In single-member settings, the incentives facing individual legislators are similar to those facing their political parties.
Table 2 presents the descriptive statistics for the natural logarithm of the total volume of public investment expenditures (Inv) and the political power variables (PolitPower) by governing party (GovParty) and by change in governing party (ChGovParty). This table shows that the political power of the Liberal Party (%ND – Votes (Seats)) was higher when the governing party changed than when the governing party was re-elected. This difference is smaller for the Socialist Party (%PASOK – Votes (Seats)). The political power of the minority parties (%Rest – Votes (Seats)) was higher when the governing party was the Liberal Party or the governing party had been re-elected. This is possibly because the minority parties comprise chiefly the two Communist Parties (the more traditional KKE and the Eurocommunist KKE-Interior and their successive transformations), which are closer to the Socialists than to the Liberals. Both the absolute and the relative political power of the governing party (%GovParty – Votes (Seats) and Dif%GovParty – Votes (Seats)) were higher when the governing party was the Liberal Party or the governing party had been re-elected. Moreover, the absolute and the relative regional political power of the governing party was higher based on the percentage of seats than on the percentage of votes. Finally, the political power of single-seat constituencies was higher when the governing party had been re-elected.
Insert Table 2 around here
The control variables (Control) include region-specific characteristics that either change over time or are time-invariant. The reason for the choice of control variables was related to their presence in the literature and previous empirical studies about pork-barrelling and corruption, as well as to data availability issues. The time-variant region-specific characteristics are GDP
per capita (divided by 10,000), as a measure of regional economic development; Density
which uses regional population density (population/km2) in order to capture regional agglomeration; and Earthquakes, which is a dummy variable coding earthquakes with victims or casualties.5 Size (in km2 divided by 10,000) is an indicator of the size (area) of the region and is a time-invariant, region-specific characteristic.6 The empirical specification also includes political period dummies to control for all political-period-specific region-invariant variables (Rodríguez-Pose, Psycharis, and Tselios 2012). All these controls capture some regional features and are likely to deal with some sources of heterogeneity reducing the omitted variable bias.
The correlations between the GovParty, ChGovParty, and Control variables are low, but the correlations between the regional political power variables (PolitPower) are high.7 This prevents the examination of the impact of all political power variables on public investment expenditures simultaneously.
5
The reason for controlling for earthquakes is that Greece is a seismic country. Past earthquakes have resulted in understandably significant changes in the allocation of public investment expenditures. Examples of such changes were the concentration of public funds in affected areas in such as Thessaloniki (1978), Korinthia (1981), Messini (1986), Achaia (1995), and Attiki (1999), after earthquakes.
6 GDP and population are not included as control variables, since they are highly correlated with each other
(0.9725) as well as with population density (0.9562 and 0.9818, respectively).
9
This paper also considers possible differences in the magnitude of the regional political
influence across different values. In other words, it examines the non-linear effects of politics
on public investment.
1. Based on data on the percentage of votes, possible differences between low, medium, and high regional political power of the Liberal Party, the Socialist Party, the minority parties, and the governing party respectively are examined.8 The high political power regions capture ‘safe’ regions; the medium political power regions capture ‘marginal’ (or ‘swing’) regions; while the so-called low political power regions capture ‘hopeless’ regions for the political parties. More specifically:
if %ND – Votes < 30, %PASOK – Votes < 30, %Rest – Votes < 10 and %GovParty –
Votes < 30, the regional political power of the Liberal Party, the Socialist Party, or
minority parties, and of the governing party, respectively, is low;
if 30 ≤ %ND – Votes ≤ 50, 30 ≤ %PASOK – Votes ≤ 50, 10 ≤ %Rest – Votes ≤ 30 and
30 ≤ %GovParty – Votes ≤ 50, the regional political power of the Liberal Party, the
Socialist Party, minority parties, and the governing party, respectively, is medium (base category), and;
if %ND – Votes > 50, %PASOK – Votes > 50, %Rest – Votes > 30 and %GovParty –
Votes > 50, the regional political power of the Liberal Party, the Socialist Party,
minority parties, and the governing party, respectively, is high.
Possible differences between a low, medium, and high relative regional political power of the governing party are also taken into account. Thus,
if Dif%GovParty – Votes < -5, the relative regional political power of the governing party is low and represents the opposition groups (i.e. ‘hopeless’ regions for the governing party);
if -5 ≤ Dif%GovParty – Votes ≤ 5, the relative regional political power of the governing party is medium (base category) and represents the marginal constituencies [i.e. ‘marginal’ (or ‘swing’) regions]; and,
if Dif%GovParty – Votes > 5, the relative regional political power of the governing party is high and represents the ruling party’s core constituencies (i.e. ‘safe’ regions for the governing party).
2. Based on data on the percentage of seats, possible differences between monopolistic, high,
medium, low, and no political power of the Liberal Party, the Socialist Party and the
governing party are considered. Hence,
if %ND – Seats = 100, %PASOK – Seats = 100, %GovParty – Seats = 100 and
Dif%GovParty - Seats = 100, the regional political power of the Liberal Party, the
Socialist Party, and the governing party (absolute or relative) respectively is monopolistic;
if 50 < %ND – Seats < 100, 50 < %PASOK – Seats < 100, 50 < %GovParty – Seats
< 100 and 0 < Dif%GovParty – Seats < 100, the regional political power of the
8 The scheme used for the allocation of the regions to the different classifications (low, medium and high) is a
combination of an ‘exogenous’ scheme, defined by criteria external to the distribution of data (e.g. regions with high political power are dominated by the two main political parties in Greece); an ‘arbitrary’ scheme, in which class boundaries are set by arbitrary criteria such as equal intervals; and an ‘ideographic’ scheme, where class boundaries are defined by the shape of the distribution (Rodríguez-Pose and Tselios 2011). Nevertheless, the regression results are robust to alternative classifications (e.g. 20 ≤ %ND (%PASOK) – Votes ≤ 40). The robustness of the regression results using alternative class boundaries and the distribution of the political variables can be provided by the authors upon request.
10
Liberal Party is higher than that of the Socialist Party, the regional political power of the Socialist Party is higher than that of the Liberal Party, and the regional political power of the governing party (absolute or relative) is higher than that of the opposition party, respectively;
if %ND – Seats = 50, %PASOK – Seats = 50, %GovParty – Seats = 50 and
Dif%GovParty – Seats = 0, the regional political power of the Liberal Party matches
that of the Socialist Party (i.e. 50 per cent) (base category);
if 0 < %ND – Seats < 50, 0 < %PASOK – Seats < 50, 0 < %GovParty – Seats < 50 and -100 < Dif%GovParty – Seats < 0, the regional political power of the Liberal Party is lower than that of the Socialist Party, the regional political power of the Socialist Party is lower than that of the Liberal Party, and the regional political power of the governing party (absolute or relative) is lower than that of the opposition party, respectively; and
if %ND – Seats = 0, %PASOK – Seats = 0, %GovParty – Seats = 0 and
Dif%GovParty - Seats = -100, there is no political power for the Liberal Party, the
Socialist Party, and the governing party (absolute or relative), respectively.
As for the minority parties, the paper analyses possible differences in political influence between a low percentage seats in the region in favour of the minority parties (%Rest - Seats
< 5) (base category) and a high percentage of seats (5 ≤ %Rest - Seats ≤ 100).
4. Regression results
This section presents the regression results of the political determinants of the regional distribution of public investment expenditures for 51 NUTS III Greek regions over 10 political periods. As the regression results do not indicate great differences in the findings between the percentage of votes and the percentage of seats data, we present only the votes data. The seats data are made available in an on-line appendix. In Tables 3-4, Regressions 1-2 and 11-11-2 test Hypothesis 1 (equation 1), Regressions 3-6 and 13-16 test Hypothesis 1-2 (equation 2), and Regressions 7-10 and 17-20 test Hypothesis 3 (equation 3). These tables show the differences in public investment expenditures between the Liberal and the Socialist Party (see coefficients on the GovParty variable) as well as between the change and no-change in the governing party (see coefficients on the ChGovParty variable). As for the interpretation of the PolitPower variable which captures pork-barrel politics, we present both the linear and non-linear effects. More specifically, Table 3 presents the regional political power of the Liberal Party (%ND – Votes) and the Socialist Party (%PASOK – Votes), and Table 4 the absolute regional political power of the governing party (%GovParty – Votes) and the relative regional political power of the governing party (Dif%GovParty – Votes).
All estimations have a good fit, as they explain more than two thirds of the variation in regional public investment expenditures.9 Because of space constraints, we do not report the coefficients on the Control variables (i.e. GDP per capita, Size, Density, and Earthquakes).10 The results on the controls show that public investment was higher in large, high-density regions and, as expected, in regions which had been affected by earthquakes causing victims or casualties. There is, however, no evidence of an association between regional economic development and regional public investment expenditures.
9 In our empirical specifications, any spillover effects may occur only within NUTS III regions, because the
Moran’s I test (Cliff and Ord 1981) adapted to the estimated residuals rejects the null hypothesis of spatial dependence.
11
The results also indicate that the total volume of public investment expenditures was higher when the Socialists were in office – during the periods 1982-1985, 1986-1989, 1994-1996, 1997-1999, and 2000-2003 – than when Greece was governed by the Liberals (see the coefficients on the GovParty variable). They also demonstrate that, after being re-elected (during the periods 1986-1989, 1997-1999, and 2000-2003), the tendency of the Socialist Party to make use of public investment for political purposes was significantly greater than when elected for the first time (see the coefficients on the ChGovParty variable).
4.1 The regional political power of the Liberal and the Socialist Party
Table 3 shows a negative association between the regional political power of the Liberal Party (%ND – Votes) and public investment in regions where the Liberals were hegemonic (%ND > 50) (Regressions 2, 4, 6, and 8). By contrast, there is a positive association between both factors in those regions where the Liberals had poor electoral results (%ND < 30) (Regressions 2, 6, and 10), and especially in periods when the Socialist Party was in office (Regression 4 vs 6). This indicates that the Liberal Party had a preference to use pork-barrel politics in order to possibly either attract opposition voters – mainly Socialist – in swing districts or with the intention of mobilising their own supporters there. The Socialists, in contrast, seem to have adopted a different approach. The results show a positive linear association between the regional political power of the Socialist Party (%PASOK – Votes) and public investment (Regressions 11, 13, 15, 17, and 19). Investment was particularly targeted at regions with a high percentage of Socialist vote (%PASOK > 50) (Regressions 12, 16, 18, and 20). As can be expected, the association was stronger when the Socialist Party was in office (Regression 14 vs 16) and, especially, after being re-elected (Regression 18 vs 20). Hence, the Socialist Party adopted a different stance with respect to pork-barrelling than the Liberals. Rather than courting opposition voters in swing constituencies, they groomed their electoral bases in safe districts.
Insert Table 3 around here
As they were never in office during the period of analysis, it comes as no surprise that there is no evidence that the regional political power of minority parties (%Rest – Votes) was translated into any changes in public investment (see Appendix 1). The regional political power of the Socialist Party was reflected in a more favourable treatment of its safe-seat constituencies in terms of public investment. There is less evidence of such behaviour in the case of the Liberal Party (see Appendix 2).
4.2 The absolute and relative regional political power of the governing party
Table 4 shows that there is no evidence of a linear statistically significant association between the absolute regional political power of the governing party and regional expenditures (Regression 1). This is possibly the result of a negative linear relationship when the Liberal Party is considered (Regressions 1, 3, 5, 7, and 9 of Table 3) and a positive one in the case of the Socialist Party (Regressions 11, 13, 15, 17, and 19 of Table 3). However, both parties seem to have behaved in a similar way after re-elections (ND in 1978-1981, PASOK in 1986-1989, PASOK in 1997-1999 and 2000-2003, and ND in 2008-2009). Re-elections of an incumbent party are connected to a greater tendency to resort to pork-barrelling, mostly
12
favouring regions that have strongly supported the party in government (%GovParty > 50) (Regressions 9 and 10). Therefore, we find that the re-elected governing parties repeatedly resorted to skewing the distribution of investments in favour of their core constituencies (i.e. ‘safe’ regions). Regions that traditionally gave a higher percentage of the vote to the re-elected governing party generally received larger shares of public investment than their neighbours that voted for the opposition party.
Insert Table 4 around here
Table 4 also depicts a positive linear relationship between the political power of the governing party relative to that of the main opposition party (Dif%GovParty – Votes) and overall public investment between 1975 and 2009 (Regression 11). Running the same regression by governing party (ND vs PASOK), there is a negative association between the relative regional political power of the Liberal Party when in office and public investment (Regression 13). Public investment tends to be higher in Liberal Party ‘safe’ regions (Dif%GovParty > 5) (Regression 14). However, an increase in the difference in votes between the Liberal Party (as governing party) and the Socialist Party is connected to declines in public investment, particularly in the safest constituencies. This implies, once again, that the Liberal Party had less of a tendency to reward those areas of the country that provided them with their strongest electoral support. PASOK behaved in a different way, strongly channelling public funds to its core constituencies (Dif%GovParty > 5) (Regression 15). The safer the region, the greater the investment (Regression 16). Thus, an increase in the difference in the political power of the Socialist Party (as governing party) relative to the Liberal Party was translated into greater public investment expenditure. The overall outcome is positive (Regression 11), reflecting that the coefficient on the relative political power is higher for PASOK than for ND (i.e. 0.0157 > 0.0067), meaning that the Socialists made greater use of pork-barrelling. Finally, the results show that, as expected, after re-elections public investments were skewed towards areas voting for the governing party (Regression 19).
Overall, the regression results depict a widespread political culture of pork-barrelling in Greece during the period of analysis. The results fail to reject Hypothesis 1 and 3. PolitPower variables show that pork-barrelling matters for the distribution of public investment expenditures (Hypothesis 1), while the coefficients on the ChGovParty variables indicate that there are differences in public investment expenditures between first-time and re-elected governments (Hypothesis 3). The only hypothesis that is rejected is Hypothesis 2, as the Socialist Party made a more extensive use of pork-barrelling than the Liberals when in office.
5. Conclusions
This paper has delved into the largely overlooked question of the presence and relevance of pork-barrel politics in the four decades after the restoration of democracy in Greece. The results of the analysis have confirmed the presence of widespread pork-barrel practices in the geographical distribution of public investment. Pork-barrel politics during the 35-year time-period between 1974 and 2009 have distorted regional development policy priorities, as the
13
distribution of public investment responded, to a large extent, to the governments’ self-interests.
However, while pork-barrelling was practiced by the two parties that alternated in power in the period between the restoration of democracy and the outbreak of the Great Recession, the analysis has revealed that each party followed a different strategy. The Liberal Party mainly allocated public investment with the aim of gaining additional support and new seats in Parliament by courting the electorate in ‘swing’ constituencies. The Socialist, by contrast, preferred rewarding electoral trust, overspending in ‘safe’ electoral districts. Re-election exacerbated these practices. Re-elected governments splashed more on pork-barrelling once the confidence of the electorate was renewed and decision-makers felt more at ease with governing Greece. The Socialist party, according to the results of the analysis, made greater public investment spending than the Liberals, as the public investment link to votes and MPs was highest during the periods between 1986-1989, 1997-1999, and 2000-2003, coinciding with the re-election of PASOK. These results are robust to measuring political power by votes as well as by MPs.
Political influences in the allocation of public investment can be considered as an important source of distortion for regional development policy priorities and a serious institutional flaw in the functioning of the Greek democracy. They may have also been a reason behind the excessive spending which led to the generation of substantial annual deficits and debts. In this respect, this article sets a marker about the role played by political parties in undermining the efficiency of the regional development effort for real or perceived party-political gains. However, this article has mostly only scratched the surface, establishing connections, but not exploring the mechanisms through which this pork-barrel was generated, allowed, managed, and tolerated. More research is needed, first, in order to assess whether Greece represents a particular case in terms of pork-barrel politics or whether this type of behaviour is – as implied by some of the literature highlighted in the literature review – more widespread, especially across other southern European countries. In addition, more qualitative and in-depth research will have to be conducted in order to assess the exact ways in which decision-makers influenced and shaped the allocation of public investment for the benefit of their parties, often at the expense of sound and efficient policies and of the Greek people as a whole.
14 Acknowledgements: We are indebted to the editor in charge, Jessie Poon, for her meticulous
handling of the different versions of the manuscript. We have also benefitted from the constructive comments of three reviewers. We thank Panagiotis Pantazis, MSc in Planning and Thematic Cartography, for putting together the map. The research leading to this article has been made possible by the generous funding of the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC [grant agreement number 269868]. Additional support was provided by the European Union (European Social Fund) and Greek national funds through the Operational Programme ‘Education and Lifelong Learning’ – Research Funding Program: THALES’ [grant number MIS 380421].
15 References
Ansolabehere, S. and J.M. Snyder. 2006. Party control of state government and the distribution of public expenditures. Scandinavian Journal of Economics 108:547-569.
Bennett, R.J. 1980. The geography of public finance: welfare under fiscal federalism and local government
finance. New York: Methuen.
Buchanan, J.M. 1950. Federalism and Fiscal Equity. American Economic Review 40:583-599.
Celbis, M.G.; D. de Crombrugghe; and J. Muysken. 2014. Public investment and regional politics: The case of Turkey: UNU-MERIT Working Papers, Maastricht Economic and social Research Institute on Innovation and Technology, Maastricht Graduate School of Governance, working papers wp2014-020.
Chen, J. 2010. The effect of electoral geography on pork barreling in bicameral legislatures. American
Journal of Political Science 54:301-322.
Cliff, A.D. and J.K. Ord. 1981. Spatial processes: models & applications. London: Pion.
Costa-i-Font, J.; E. Rodriguez-Oreggia; and D. Lunapla. 2003. Political competition and pork-barrel politics in the allocation of public investment in Mexico. Public Choice 116:185-204.
Cox, G.W. and M.D. McCubbins. 1986. Electoral politics as a redistributive game. Journal of Politics
48:370-389.
Cutts, D. and D.J. Webber. 2010. Voting patterns, party spending and relative location in England and Wales. Regional Studies 44:735-760.
Dahlberg, M. and E. Johansson. 2002. On the vote-purchasing behavior of incumbent governments.
American Political Science Review 96:27-40.
De La Calle, L. and L. Orriols. 2010. Explaining the electoral effects of public investments: The case of the expansion of the underground in Madrid, 1995-2007. European Journal of Political Research
49:393-417.
Dixit, A. and J. Londregan. 1996. The determinants of success of special interests in redistributive politics.
Journal of Politics 58:1132-1155.
European Commission. 2014. Report from the Commission to the Council and the European Parliament: EU anti-corruption report: Brussels, 3.2.2014; COM(2014) 38 final.
Golden, M. and B. Min. 2013. Distributive politics around the world. Annual Review of Political Science,
Vol 16 16:73-+.
Golden, M.A. and L. Picci. 2008. Pork-barrel politics in postwar Italy, 1953-94. American Journal of
Political Science 52:268-289.
Johnston, R.J. 1977. Environment, elections and expenditure: analyses of where governments spend.
Regional Studies 11:383-394.
Lambrinidis, M.; Y. Psycharis; and A. Rovolis. 2005. Regional allocation of public infrastructure investment: the case of Greece. Regional Studies 39:1231-1244.
Limosani, M. and P. Navarra. 2001. Local pork-barrel politics in national pre-election dates: The case of Italy. Public Choice 106:317-326.
Lindbeck, A. and J.W. Weibull. 1987. Balanced-budget redistribution as the outcome of political competition. Public Choice 52:273-297.
Luca, D. and A. Rodríguez-Pose. 2015. Electoral politics and regional development: assessing the role of equity, efficiency, and politics in the geography of Turkey’s public investment. Journal of
Development Studies 51(11): 1518-1540.
Oates, W.E. 1972. Fiscal federalism. New York: Harcourt Brace Jovanovich.
Primo, D.M. and J.M. Snyder, Jr. 2010. Party Strength, the Personal Vote, and Government Spending.
American Journal of Political Science 54:354-370.
Psycharis, Y. 1990. Fiscal Crisis and Regional Development: Unpublished PhD Dissertation, Panteion University, Athens.
———. 2008. Public spending patterns: The regional allocation of public investment in Greece by political period: GreeSE Paper No 14, Hellenic Observatory Papers on Greece and Southeast Europe. Rodríguez-Pose, A.; Y. Psycharis; and V. Tselios. 2012. Public investment and regional growth and
convergence: Evidence from Greece. Papers in Regional Science 91:543-568.
———. 2015. Politics and Investment: Examining the Territorial Allocation of Public Investment in Greece. Regional Studies forthcoming (http://dx.doi.org/10.1080/00343404.2015.1009027).
16 Rodríguez-Pose, A. and V. Tselios. 2011. Mapping the European regional educational distribution.
European Urban and Regional Studies 18:358-374.
Stokes, S.C. 2005. Perverse accountability: A formal model of machine politics with evidence from Argentina. American Political Science Review 99:315-325.
Weghorst, K.R. and S.I. Lindberg. 2013. What drives the swing voter in Africa? American Journal of
17 Table 1: Political periods
a/a Date of national
elections
Political period Governing party Change or no change
in governing party
1 17-Nov-1974 1975-1977 ND Change
2 20-Nov-1977 1978-1981 ND No Change
3 18-Oct-1981 1982-1985 PASOK Change
4 02-Jun-1985 1986-1989 PASOK No Change
5 08-Apr-1990 1990-1993 ND Change
6 10-Oct-1993 1994-1996 PASOK Change
7 22-Sept-1996 1997-1999 PASOK No Change
8 09-Apr-2000 2000-2003 PASOK No Change
9 07-Mar-2004 2004-2007 ND Change
10 16-Sept-2007 2008-2009 ND No Change
18 Table 2: Descriptive statistics
Variable
GovParty or
ChGovParty Obs Mean Std. Dev. Min Max Inv ND 255 18.2680 0.8864 16.3872 21.9577 PASOK 255 18.4727 0.7714 16.9068 22.2913 CHANGE 255 18.3937 0.8419 16.3872 21.9577 NO CHANGE 255 18.3470 0.8319 16.7967 22.2913 %ND – Votes ND 255 47.4158 8.5953 15.2608 76.0418 PASOK 255 41.0887 5.9307 25.1183 56.9166 CHANGE 255 46.0856 9.1732 15.2608 76.0418 NO CHANGE 255 42.4188 6.1848 16.0087 56.9166 %PASOK – Votes ND 255 33.6072 10.6966 10.8753 56.5183 PASOK 255 45.4700 5.9932 30.4159 64.3227 CHANGE 255 39.6439 11.5178 11.8679 64.3227 NO CHANGE 255 39.4333 9.3965 10.8753 64.3227 %Rest – Votes ND 255 18.9770 10.7596 0.6593 60.5497 PASOK 255 13.4413 5.2459 4.8921 40.4836 CHANGE 255 14.2706 7.5967 0.6593 46.8271 NO CHANGE 255 18.1478 9.6652 5.5727 60.5497 %GovParty – Votes ND 255 47.4158 8.5953 15.2608 76.0418 PASOK 255 45.4700 5.9932 30.4159 64.3227 CHANGE 255 49.0039 7.8962 15.2608 76.0418 NO CHANGE 255 43.8818 6.0148 16.0087 64.3227 Dif%GovParty – Votes ND 255 13.8085 16.1500 -22.6513 63.1174 PASOK 255 4.3814 10.7081 -22.0465 38.3043 CHANGE 255 12.2784 16.3159 -22.6513 63.1174 NO CHANGE 255 5.9115 11.5559 -22.0465 38.7679 %ND – Seats ND 255 64.1391 23.4825 0 100 PASOK 255 38.4885 20.1750 0 100 CHANGE 255 53.9657 25.2271 0 100 NO CHANGE 255 48.6619 25.2627 0 100 %PASOK – Seats ND 255 30.1332 20.3458 0 100 PASOK 255 59.0030 19.2656 0 100 CHANGE 255 42.1030 24.6845 0 100 NO CHANGE 255 47.0333 24.1239 0 100 %Rest – Seats ND 255 5.7278 14.9495 0 100 PASOK 255 2.5085 6.4278 0 33.33334 CHANGE 255 3.9314 13.6507 0 100 NO CHANGE 255 4.3048 9.1430 0 50 %GovParty – Seats ND 255 64.1391 23.4825 0 100 PASOK 255 59.0030 19.2656 0 100 CHANGE 255 62.8678 21.5378 0 100 NO CHANGE 255 60.2743 21.6467 0 100 Dif%GovParty – Seats ND 255 34.0059 41.3191 -100 100 PASOK 255 20.5145 38.9239 -100 100 CHANGE 255 29.6670 39.5323 -100 100 NO CHANGE 255 24.8534 41.7046 -100 100 Single-seat constituency ND 255 0.1137 0.3181 0 1 PASOK 255 0.1059 0.3083 0 1 CHANGE 255 0.0980 0.2980 0 1 NO CHANGE 255 0.1216 0.3274 0 1
19 Table 3: The regional political power of the Liberal Party (%ND – Votes) and the Socialist Party (%PASOK – Votes)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ALL ND ND PASOK PASOK CHANGE CHANGE NO
CHANGE
NO CHANGE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
a. GovParty
PASOK base base base base base base
ND -0.6052*** -0.6231*** -0.1681 -0.2291** -0.6573*** -0.6391***
b. ChGovParty
Change base base base base base base
No change 0.0484 0.0891 0.1193 0.1307 0.2488*** 0.1710**
c. PolitPower Political power of the Liberal Party
%ND – Votes -0.0161*** -0.0097** -0.0275*** -0.0154*** -0.0170***
%ND < 30 0.3010** 0.2261 0.3767** 0.2456 0.3745*
30 ≤ %ND ≤ 50 base base base base base
%ND > 50 -0.1942*** -0.1412* -0.4034*** -0.2180*** -0.1565
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7302 0.7255 0.7122 0.7124 0.7577 0.7417 0.7470 0.7440 0.7141 0.7084
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20)
a. GovParty
PASOK base base base base base base
ND -0.5234*** -0.5812*** -0.1059 -0.5059*** -0.1401 -0.2571
b. ChGovParty
Change base base base base base base
No change -0.0724 0.1119 0.0378 0.0994 0.2027** 0.1694**
c. PolitPower Political power of the Socialist Party
%PASOK – Votes 0.0172*** 0.0105* 0.0250*** 0.0114** 0.0249***
%PASOK < 30 -0.1278 -0.1401 0.1611 -0.3885**
30 ≤ %PASOK ≤ 50 base base base base base
%PASOK > 50 0.2906*** 0.1219 0.3349*** 0.2153** 0.3855***
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7295 0.7274 0.7111 0.7086 0.7537 0.7475 0.7387 0.7393 0.7252 0.7234
20 Table 4: The absolute and relative regional political power of the governing party (%GovParty – Votes and Dif%GovParty – Votes, respectively)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ALL ND ND PASOK PASOK CHANGE CHANGE NO
CHANGE
NO CHANGE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
a. GovParty
GovParty: PASOK Base base base base base base
GovParty: ND -0.6159*** -0.6115*** -0.3859*** -0.3949*** -0.5836*** -0.6390***
b. ChGovParty
ChGovParty: Change Base base
ChGovParty: No change 0.2665** 0.2989***
c. PolitPower
Absolute political power of the governing party
%GovParty – Votes 0.0046 -0.0012 0.0127**
%GovParty < 30 0.3092 0.2259 0.3267
30 ≤ %GovParty ≤ 50 base base base
%GovParty > 50 0.0995* -0.0162 0.2873***
Observations 510 510 255 255 255 255
R-squared 0.7172 0.7193 0.7320 0.7326 0.7088 0.7155
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20)
a. GovParty
GovParty: PASOK Base base base base base base
GovParty: ND -0.6334*** -0.6363*** -0.4309*** -0.4057*** -0.7401*** -0.6855***
b. ChGovParty
ChGovParty: Change Base base base base base base
ChGovParty: No change 0.3290*** 0.2336** 0.0328 0.1953 0.2516*** 0.2074**
c. PolitPower
Relative political power of the governing party
Dif%GovParty – Votes 0.0038** -0.0067** 0.0157*** 0.0012 0.0074**
Dif%GovParty < -5 -0.0285 -0.2967** 0.1729** -0.0799 0.0083
-5 ≤ Dif%GovParty ≤ 5 base base base base base
Dif%GovParty > 5 0.0440 -0.3453*** 0.3413*** -0.0561 0.1307
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7184 0.7173 0.7134 0.7183 0.7624 0.7423 0.7322 0.7325 0.7085 0.7056
*** p<0.01, ** p<0.05, * p<0.1; all regressions include controls, constant and time-dummies; we do not present Reg. 3 and 4 because they are presented in Reg. 3 and 4 of Table 3 and, similarly, we do not present Reg. 5 and 6 because they are presented in Reg. 15 and 16 of Table 3.
21 Figure 1: The regional distribution of public investment expenditures
22 Appendix 1: The regional political power of the minority parties (%Rest – Votes)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ALL ND ND PASOK PASOK CHANGE CHANGE NO
CHANGE
NO CHANGE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
a. GovParty
GovParty: PASOK base base base base base base
GovParty: ND -0.6270*** -0.6444*** -0.4797*** -0.4422*** -0.4856*** -0.5527***
b. ChGovParty
ChGovParty: Change base base base base base base
ChGovParty: No change 0.2242** 0.2148** 0.2402* 0.2355* 0.0895 0.0885
c. PolitPower Political power of the minority parties
%Rest – Votes 0.0022 0.0029 0.0008 0.0079* -0.0055
%Rest < 10 -0.0893* -0.1822* -0.0331 -0.0613 -0.1172
10 ≤ %Rest ≤ 30 base base base base base
%Rest > 30 -0.0084 -0.0087 0.2972 0.1693 -0.1629
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7162 0.7177 0.7070 0.7111 0.7210 0.7220 0.7350 0.7346 0.7022 0.7050
23 Appendix 2: Comparing the regional political power of the Liberal Party (%ND – Votes), the Socialist Party (%PASOK – Votes) and the minority parties (%Rest – Votes)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ND PASOK CHANGE NO
CHANGE
(1) (2) (3) (4) (5)
a. GovParty
GovParty: PASOK base base base
GovParty: ND -0.5517*** -0.1255 -0.2249
b. ChGovParty
ChGovParty: Change base base base
ChGovParty: No change -0.0704 0.0372 0.2555***
c. PolitPower
Comparing the political power
%ND – Votes base base base base base
%PASOK – Votes 0.0212*** 0.0134** 0.0315*** 0.0165*** 0.0273***
%Rest – Votes 0.0106*** 0.0071 0.0184*** 0.0141*** 0.0060
Observations 510 255 255 255 255
R-squared 0.7337 0.7134 0.7627 0.7472 0.7264
24 On-line Appendix 1: The regional political power of the Liberal Party (%ND – Seats) and the Socialist Party (%PASOK – Seats)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ALL ND ND PASOK PASOK CHANGE CHANGE NO
CHANGE
NO CHANGE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
a. GovParty
GovParty: PASOK base base base base base base
GovParty: ND -0.5514*** -0.5694*** -0.2247** -0.2221* -0.5216*** -0.4685***
b. ChGovParty
ChGovParty: Change base base base base base base
ChGovParty: No change 0.1289 0.0632 0.1436 0.0910 0.0974 0.1493*
c. PolitPower Political power of the Liberal Party
%ND – Seats -0.0035*** -0.0031** -0.0036*** -0.0038** -0.0032**
%ND = 0 -0.1036 -0.1909 -0.1200 -0.0988 -0.1089
0 < %ND < 50 0.2071*** 0.2407** 0.2002*** 0.1761** 0.2253***
%ND = 50 base base base base base
50 < %ND < 100 -0.0317 0.0395 -0.1487* 0.0191 -0.0646 %ND = 100 -0.2999*** -0.2022* -0.5036** -0.2640** -0.3250*** Observations 510 510 255 255 255 255 255 255 255 255 R-squared 0.7231 0.7361 0.7116 0.7221 0.7294 0.7539 0.7387 0.7468 0.7083 0.7272 (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) a. GovParty
GovParty: PASOK base base base base base base
GovParty: ND -0.5503*** -0.5223*** -0.2549** -0.2382** -0.5264*** -0.4289***
b. ChGovParty
ChGovParty: Change base base base base base base
ChGovParty: No change 0.1601 0.1295 0.1784 0.1446 0.1005 0.1560*
c. PolitPower Political power of the Socialist Party
%PASOK – Seats 0.0029*** 0.0022 0.0031** 0.0032** 0.0026*
%PASOK = 0 -0.2981*** -0.2064* -0.5073** -0.2194** -0.3966***
0 < %PASOK < 50 -0.0506 0.0243 -0.1846** 0.0356 -0.1335
%PASOK = 50 base base base base base
50 < %PASOK < 100 0.2030*** 0.1786 0.1931*** 0.1953** 0.1994**
%PASOK = 100 -0.1840** -0.4642 -0.2079** -0.0951 -0.2474**
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7201 0.7371 0.7087 0.7190 0.7264 0.7628 0.7366 0.7465 0.7047 0.7312
25 On-line Appendix 2: The absolute and relative regional political power of the governing party (%GovParty – Seats and Dif%GovParty – Seats,
respectively)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ALL ND ND PASOK PASOK CHANGE CHANGE NO
CHANGE
NO CHANGE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
a. GovParty
GovParty: PASOK Base base base base base base
GovParty: ND -0.6162*** -0.6290*** -0.3624*** -0.3231*** -0.6132*** -0.6158***
b. ChGovParty
ChGovParty: Change Base base
ChGovParty: No change 0.2051* 0.1482
c. PolitPower
Absolute political power of the governing party
%GovParty – Seats -0.0002 -0.0015 0.0008
%GovParty = 0 -0.3319** -0.2561 -0.3722*
0 < %GovParty < 50 0.0035 0.1290 -0.0738
%GovParty = 50 base base base
50 < %GovParty < 100 0.1218** 0.0950 0.1493** %GovParty = 100 -0.1974*** -0.1864* -0.2147** Observations 510 510 255 255 255 255 R-squared 0.7160 0.7321 0.7330 0.7445 0.7014 0.7250 (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) a. GovParty
GovParty: PASOK Base base base base base base
GovParty: ND -0.6182*** -0.6304*** -0.3670*** -0.3260*** -0.6251*** -0.6318***
b. ChGovParty
ChGovParty: Change Base base base base base base
ChGovParty: No change 0.2186** 0.1357 0.1521 0.1260 0.0995 0.1464*
c. PolitPower
Relative political power of the governing party
Dif%GovParty – Seats 0.0002 -0.0016* 0.0017** -0.0007 0.0008
Dif%GovParty = -100 -0.4990*** -0.4683 -0.5057** -0.3286 -0.5580***
-100 < Dif%GovParty < 0 -0.0123 0.2842** -0.1476* 0.1370 -0.0970
Dif%GovParty = 0 base base base base base
0 < Dif%GovParty < 100 0.1047** -0.0316 0.2104*** 0.0812 0.1306*
Dif%GovParty = 100 -0.2091*** -0.2396** -0.2047** -0.1888* -0.2337**
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7161 0.7321 0.7106 0.7243 0.7280 0.7610 0.7326 0.7429 0.7025 0.7268
*** p<0.01, ** p<0.05, * p<0.1; all regressions include controls, constant and time-dummies; we do not present Reg. 3 and 4 because they are presented in Reg. 3 and 4 of on-line Appendix 1 and, similarly, we do not present Reg. 5 and 6 because they are presented in Reg. 15 and 16 of on-line Appendix 1.
26 On-line Appendix 3: The political power of the single-seat constituency
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ND PASOK CHANGE NO
CHANGE
(1) (2) (3) (4) (5)
a. GovParty
GovParty: PASOK base base base
GovParty: ND -0.6074*** -0.4050*** -0.6083***
b. ChGovParty
ChGovParty: Change base base base
ChGovParty: No change 0.2335** 0.2310* 0.1072
c. PolitPower
Single-seat constituency -0.2586*** -0.1889* -0.3337*** -0.1979* -0.3082***
Observations 510 255 255 255 255
R-squared 0.7236 0.7102 0.7354 0.7359 0.7129
27 On-line Appendix 4: The regional political power of the minority parties (%Rest – Seats)
Hypothesis 1 Hypothesis 2 Hypothesis 3
ALL ALL ND ND PASOK PASOK CHANGE CHANGE NO
CHANGE
NO CHANGE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
a. GovParty
GovParty: PASOK base base base base base base
GovParty: ND -0.6320*** -0.6583*** -0.3949*** -0.3869*** -0.6439*** -0.6404***
b. ChGovParty
ChGovParty: Change base base base base base Base
ChGovParty: No change 0.1895* 0.1567 0.2040* 0.1685 0.0736 0.0686
c. PolitPower Political power of the minority parties
%Rest – Seats 0.0035* 0.0027 0.0106** 0.0015 0.0086**
%Rest < 5 base base base base base
5 ≤ %Rest ≤ 100 0.2619*** 0.2598*** 0.2718*** 0.2730*** 0.2505***
Observations 510 510 255 255 255 255 255 255 255 255
R-squared 0.7181 0.7271 0.7083 0.7182 0.7273 0.7322 0.7325 0.7426 0.7082 0.7120
